A Fully Automatic Technique for Precise Localization and Quantification of Amyloid-β PET Scans

Area of Science:

• Neuroimaging
• Neuropathology
• Alzheimer Disease Research

Background:

• Alzheimer disease (AD) is characterized by amyloid-β (Aβ) plaque accumulation, exhibiting spatial heterogeneity across the brain.
• Accurate localization and quantification of Aβ pathology are crucial for understanding AD progression and developing effective treatments.
• Positron Emission Tomography (PET) imaging is a key tool for in vivo assessment of Aβ burden.

Purpose of the Study:

• To develop and validate a novel PET imaging technique for high-precision, millimeter-scale quantification of Aβ pathology in the brain.
• To enhance the accuracy of Standardized Uptake Value Ratio (SUVR) measurements for Aβ PET imaging.
• To improve the correlation between PET-derived Aβ quantification and postmortem histopathologic findings.

Main Methods:

• Developed a technique using intermodal and intrasubject registration with normalized mutual information to align FreeSurfer neuroanatomic labels with PET image space.
• Computed regional SUV ratio (SUVR) using transformed labels.
• Validated the technique against postmortem histopathologic staining data from 52 older participants.

Main Results:

• The novel method yielded consistently and significantly higher SUVRs compared to conventional methods across most regions of interest.
• A 2-way ANOVA confirmed a significant main effect of the method and a significant interaction effect on the relationship between computed SUVR and histopathologic staining scores.
• The improved SUVRs demonstrated a closer association with histopathologic data.

Conclusions:

• Processing Aβ PET data in native space enhances the accuracy of SUVR measurements.
• This advanced technique provides a more precise assessment of Aβ pathology, closely aligning with histopathologic gold standards.
• The findings support the utility of this method for improved diagnosis and monitoring of Alzheimer disease.